Recommender systems have become an integral part of online platforms. Every day the volume of training data is expanding and the number of user interactions is constantly increasing. The exploration of larger and more expressive models has become a necessary pursuit to improve user experience. However, this progression carries with it an increased computational burden. In commercial settings, once a recommendation system model has been trained and deployed it typically needs to be updated frequently as new client data arrive. Cumulatively, the mounting volume of data is guaranteed to eventually make full batch retraining of the model from scratch computationally infeasible. Naively fine-tuning solely on the new data runs into the well-documented problem of catastrophic forgetting. Despite the fact that negative sampling is a crucial part of training with implicit feedback, no specialized technique exists that is tailored to the incremental learning framework. In this work, we take the first step to propose, a personalized negative reservoir strategy which is used to obtain negative samples for the standard triplet loss. This technique balances alleviation of forgetting with plasticity by encouraging the model to remember stable user preferences and selectively forget when user interests change. We derive the mathematical formulation of a negative sampler to populate and update the reservoir. We integrate our design in three SOTA and commonly used incremental recommendation models. We show that these concrete realizations of our negative reservoir framework achieve state-of-the-art results in standard benchmarks, on multiple standard top-k evaluation metrics.
Temporal Graph Neural Networks have garnered substantial attention for their capacity to model evolving structural and temporal patterns while exhibiting impressive performance. However, it is known that these architectures are encumbered by issues that constrain their performance, such as over-squashing and over-smoothing. Meanwhile, Transformers have demonstrated exceptional computational capacity to effectively address challenges related to long-range dependencies. Consequently, we introduce Todyformer-a novel Transformer-based neural network tailored for dynamic graphs. It unifies the local encoding capacity of Message-Passing Neural Networks (MPNNs) with the global encoding of Transformers through i) a novel patchifying paradigm for dynamic graphs to improve over-squashing, ii) a structure-aware parametric tokenization strategy leveraging MPNNs, iii) a Transformer with temporal positional-encoding to capture long-range dependencies, and iv) an encoding architecture that alternates between local and global contextualization, mitigating over-smoothing in MPNNs. Experimental evaluations on public benchmark datasets demonstrate that Todyformer consistently outperforms the state-of-the-art methods for downstream tasks. Furthermore, we illustrate the underlying aspects of the proposed model in effectively capturing extensive temporal dependencies in dynamic graphs.
Despite Multi-modal Large Language Models (MM-LLMs) have made exciting strides recently, they are still struggling to efficiently model the interactions among multi-modal inputs and the generation in non-textual modalities. In this work, we propose TEAL (Tokenize and Embed ALl)}, an approach to treat the input from any modality as a token sequence and learn a joint embedding space for all modalities. Specifically, for the input from any modality, TEAL first discretizes it into a token sequence with the off-the-shelf tokenizer and embeds the token sequence into a joint embedding space with a learnable embedding matrix. MM-LLMs just need to predict the multi-modal tokens autoregressively as the textual LLMs do. Finally, the corresponding de-tokenizer is applied to generate the output in each modality based on the predicted token sequence. With the joint embedding space, TEAL enables the frozen LLMs to perform both understanding and generation tasks involving non-textual modalities, such as image and audio. Thus, the textual LLM can just work as an interface and maintain its high performance in textual understanding and generation. Experiments show that TEAL achieves substantial improvements in multi-modal understanding, and implements a simple scheme for multi-modal generations.
The performance of transformers for time-series forecasting has improved significantly. Recent architectures learn complex temporal patterns by segmenting a time-series into patches and using the patches as tokens. The patch size controls the ability of transformers to learn the temporal patterns at different frequencies: shorter patches are effective for learning localized, high-frequency patterns, whereas mining long-term seasonalities and trends requires longer patches. Inspired by this observation, we propose a novel framework, Multi-resolution Time-Series Transformer (MTST), which consists of a multi-branch architecture for simultaneous modeling of diverse temporal patterns at different resolutions. In contrast to many existing time-series transformers, we employ relative positional encoding, which is better suited for extracting periodic components at different scales. Extensive experiments on several real-world datasets demonstrate the effectiveness of MTST in comparison to state-of-the-art forecasting techniques.
Negative sampling methods are vital in implicit recommendation models as they allow us to obtain negative instances from massive unlabeled data. Most existing approaches focus on sampling hard negative samples in various ways. These studies are orthogonal to the recommendation model and implicit datasets. However, such an idea contradicts the common belief in AutoML that the model and dataset should be matched. Empirical experiments suggest that the best-performing negative sampler depends on the implicit dataset and the specific recommendation model. Hence, we propose a hypothesis that the negative sampler should align with the capacity of the recommendation models as well as the statistics of the datasets to achieve optimal performance. A mismatch between these three would result in sub-optimal outcomes. An intuitive idea to address the mismatch problem is to exhaustively select the best-performing negative sampler given the model and dataset. However, such an approach is computationally expensive and time-consuming, leaving the problem unsolved. In this work, we propose the AutoSample framework that adaptively selects the best-performing negative sampler among candidates. Specifically, we propose a loss-to-instance approximation to transform the negative sampler search task into the learning task over a weighted sum, enabling end-to-end training of the model. We also designed an adaptive search algorithm to extensively and efficiently explore the search space. A specific initialization approach is also obtained to better utilize the obtained model parameters during the search stage, which is similar to curriculum learning and leads to better performance and less computation resource consumption. We evaluate the proposed framework on four benchmarks over three models. Extensive experiments demonstrate the effectiveness and efficiency of our proposed framework.
With the continuous increase of users and items, conventional recommender systems trained on static datasets can hardly adapt to changing environments. The high-throughput data requires the model to be updated in a timely manner for capturing the user interest dynamics, which leads to the emergence of streaming recommender systems. Due to the prevalence of deep learning-based recommender systems, the embedding layer is widely adopted to represent the characteristics of users, items, and other features in low-dimensional vectors. However, it has been proved that setting an identical and static embedding size is sub-optimal in terms of recommendation performance and memory cost, especially for streaming recommendations. To tackle this problem, we first rethink the streaming model update process and model the dynamic embedding size search as a bandit problem. Then, we analyze and quantify the factors that influence the optimal embedding sizes from the statistics perspective. Based on this, we propose the \textbf{D}ynamic \textbf{E}mbedding \textbf{S}ize \textbf{S}earch (\textbf{DESS}) method to minimize the embedding size selection regret on both user and item sides in a non-stationary manner. Theoretically, we obtain a sublinear regret upper bound superior to previous methods. Empirical results across two recommendation tasks on four public datasets also demonstrate that our approach can achieve better streaming recommendation performance with lower memory cost and higher time efficiency.
Recommender systems now consume large-scale data and play a significant role in improving user experience. Graph Neural Networks (GNNs) have emerged as one of the most effective recommender system models because they model the rich relational information. The ever-growing volume of data can make training GNNs prohibitively expensive. To address this, previous attempts propose to train the GNN models incrementally as new data blocks arrive. Feature and structure knowledge distillation techniques have been explored to allow the GNN model to train in a fast incremental fashion while alleviating the catastrophic forgetting problem. However, preserving the same amount of the historical information for all users is sub-optimal since it fails to take into account the dynamics of each user's change of preferences. For the users whose interests shift substantially, retaining too much of the old knowledge can overly constrain the model, preventing it from quickly adapting to the users' novel interests. In contrast, for users who have static preferences, model performance can benefit greatly from preserving as much of the user's long-term preferences as possible. In this work, we propose a novel training strategy that adaptively learns personalized imitation weights for each user to balance the contribution from the recent data and the amount of knowledge to be distilled from previous time periods. We demonstrate the effectiveness of learning imitation weights via a comparison on five diverse datasets for three state-of-art structure distillation based recommender systems. The performance shows consistent improvement over competitive incremental learning techninques.
Personalized recommender systems have been widely studied and deployed to reduce information overload and satisfy users' diverse needs. However, conventional recommendation models solely conduct a one-time training-test fashion and can hardly adapt to evolving demands, considering user preference shifts and ever-increasing users and items in the real world. To tackle such challenges, the streaming recommendation is proposed and has attracted great attention recently. Among these, continual graph learning is widely regarded as a promising approach for the streaming recommendation by academia and industry. However, existing methods either rely on the historical data replay which is often not practical under increasingly strict data regulations, or can seldom solve the \textit{over-stability} issue. To overcome these difficulties, we propose a novel \textbf{D}ynamically \textbf{E}xpandable \textbf{G}raph \textbf{C}onvolution (DEGC) algorithm from a \textit{model isolation} perspective for the streaming recommendation which is orthogonal to previous methods. Based on the motivation of disentangling outdated short-term preferences from useful long-term preferences, we design a sequence of operations including graph convolution pruning, refining, and expanding to only preserve beneficial long-term preference-related parameters and extract fresh short-term preferences. Moreover, we model the temporal user preference, which is utilized as user embedding initialization, for better capturing the individual-level preference shifts. Extensive experiments on the three most representative GCN-based recommendation models and four industrial datasets demonstrate the effectiveness and robustness of our method.
Multi-types of user behavior data (e.g., clicking, adding to cart, and purchasing) are recorded in most real-world recommendation scenarios, which can help to learn users' multi-faceted preferences. However, it is challenging to explore multi-behavior data due to the unbalanced data distribution and sparse target behavior, which lead to the inadequate modeling of high-order relations when treating multi-behavior data ''as features'' and gradient conflict in multitask learning when treating multi-behavior data ''as labels''. In this paper, we propose CIGF, a Compressed Interaction Graph based Framework, to overcome the above limitations. Specifically, we design a novel Compressed Interaction Graph Convolution Network (CIGCN) to model instance-level high-order relations explicitly. To alleviate the potential gradient conflict when treating multi-behavior data ''as labels'', we propose a Multi-Expert with Separate Input (MESI) network with separate input on the top of CIGCN for multi-task learning. Comprehensive experiments on three large-scale real-world datasets demonstrate the superiority of CIGF. Ablation studies and in-depth analysis further validate the effectiveness of our proposed model in capturing high-order relations and alleviating gradient conflict. The source code and datasets are available at https://github.com/MC-CV/CIGF.